| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ANIMAL GENETICS |
* Universidade de Sao Paulo, Cx. P. 23, 13635-970, Pirassununga, SP, Brazil;
and
Oklahoma State University, Stillwater 74078;
and
Agropecuaria CFM Ltda., Cx. P. 293, 15035-900, S. J. do Rio Preto, SP, Brazil; and
and
Colorado State University, Fort Collins 80523
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and MethodsResults and DiscussionImplicationsLiterature Cited |
|---|
in single-trait analyses were 0.68 ± 0.09 and 0.61 ± 0.10 using CG1 and CG2 definitions, respectively. Heritability estimates for two-trait analyses were 0.69 ± 0.09 (CG1) and 0.63 ± 0.08 (CG2) for heifer pregnancy and 0.57 ± 0.03 (both CG) for scrotal circumference. The genetic correlation estimates between the two traits were 0.20 ± 0.12 (CG1) and 0.20 ± 0.13 (CG2). Based on the results of this study, EPD for heifer pregnancy can be used to select bulls for the production of precocious daughters and will be more effective than selecting on scrotal circumference EPD in Nellore cattle. However, scrotal circumference can be incorporated in a two-trait analysis to increase the accuracy of prediction for heifer pregnancy EPD for young bulls. Using contemporary group without heifer weaning management gave higher h2 and, for two-trait analysis, converged more quickly.
Key Words: Expected Progeny Difference • Genetic Correlation • Heritability • Method • Nonlinear Model • Sexual Precocity
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and MethodsResults and DiscussionImplicationsLiterature Cited |
|---|
. Age at puberty in extensive management systems can be estimated from age at first calving. Pereira et al. (2002) reported an average of 3 yr for age at first calving in Nellore, and it was dependent on the heifer’s age when exposed to breeding. According to Fajersson et al. (1991), Zebu heifers in the tropics fed according to NRC protein recommendations can reach puberty at 12.3 mo and calve at 27 mo of age.
Among measures of fertility and precocity, scrotal circumference is an easy and inexpensive trait to measure and is reported to be favorably associated with age at puberty (Brinks et al., 1978; Smith et al., 1989). In Nellore, scrotal circumference has high heritability (h2) (Eler et al., 1996; Quirino and Bergman, 1998), and it is reported to be favorably associated with age at first calving (Martins Filho and Lôbo, 1991; Pereira et al., 2000).
Doyle et al. (1996) and Evans et al. (1999) evaluated heifer pregnancy, defined as the probability of a heifer being pregnant after the end of the breeding season when she is exposed to a bull or inseminated. Eler et al. (2002), using a single-trait animal model to analyze pregnancy records of Nellore heifers exposed to breeding at about 14 mo of age, reported a high h2 for heifer pregnancy. The estimated h2 reported by the authors was 0.57 ± 0.11.
The genetic relationship between heifer pregnancy and scrotal circumference is also very important, although scrotal circumference is no more than an indicator trait for age at puberty. The magnitude of its relationship with female reproduction is still not well established. Evans et al. (1999) reported an estimate close to zero for the genetic correlation between the two traits in Hereford cattle.
The main objectives of this study were to evaluate previous work reporting heifer pregnancy as a highly heritable trait using additional data and to determine whether there is any advantage of incorporating scrotal circumference data into an EPD for heifer pregnancy in Nellore.
|
Material and Methods |
|---|
|
TopAbstractIntroductionMaterial and MethodsResults and DiscussionImplicationsLiterature Cited |
|---|
Cattle Management
Both bulls and heifers were maintained on high-quality pasture (40% Brachiaria brizantha, 50% Panicum maximum, and 10% others) and provided only salt and mineral supplementation (11% Ca, 6% P, 1% Mg, 4% S, 16% Na, 0.15% Cu, 0.15% Mn, 0.45% Zn, 0.015% I, 0.007% Co, and 0.002% Se). Calves were born from late August to December and remained with their dams up to 7 mo of age on high-quality pasture. The 90-d heifer breeding season started in November and ended in January. Using multiple-sire or single-sire breeding pastures, heifers were assigned by birth date to breeding management groups with heifers of similar age. The ratio of heifers per bull was about 35:1. All heifers were evaluated for pregnancy (rectal palpation) approximately 60 d after breeding. At the start of the breeding season, all heifers in the herd were exposed at breeding regardless of their weight or body condition. Heifers that failed to conceive at 14 mo were retained until the next breeding season, when they were exposed as 2-yr-olds. At this time, they were removed from the herd if they did not conceive.
Cows older than 2 yr of age that did not conceive or had poor progeny performance were culled. Bulls were selected based on an index including EPD for weaning weight, postweaning gain, scrotal circumference, and muscle score, in a proportion of 20, 40, 20, and 20% respectively. The EPD were transformed in standard deviations. Attention was paid on age at first calving and birth weight EPD. Since 2000, scrotal circumference EPD was replaced with heifer pregnancy EPD in the index for selection purposes, but not to rank young bulls for selling.
Heifer Pregnancy Analysis
Data Description.
Heifer pregnancy was defined as the observation that a heifer conceives and remains pregnant to palpation, given that she was exposed at breeding. Pregnant heifers or heifers with a calving record were scored as 1, and nonpregnant heifers were scored as 0. The data set contained 18,145 records on heifers born from 1993 through 2001 and exposed to breeding at approximately 14 mo of age from 1994 through 2002. The heifers were daughters of 290 sires and 12,020 dams. Owing to the use of multisire pasture, sires were not known for 4,226 heifers from those groups. Of the 18,145 records, 3,062 (17%) were scored 1 and 15,080 (83.0%) were scored 0.
Pedigrees included all animals with an observation plus all known relationships up to a maximum of nine generations, resulting in a total of 37,763 animals, including 713 sires, 20,742 dams, and 19,500 base animals. For this analysis, a base animal was an individual with one or both parents not known. The number of animals with unknown sires is large for this population owing to the use of multiple-sire pasture groups.
Statistical Procedures.
Using Method procedures, variance components were estimated for heifer pregnancy with a categorical animal model (Reverter et al., 1994; Snelling et al., 1995; Evans et al., 1999). Method procedures used random 50% subsamples from the data and compared those subsamples with all the data. The number of subsamples used was such that the standard error of the mean heritability was less than 0.0099 for all analyses (Tables 1 and 2).
|
|
 |
where
 |
and Y = vector of "pseudo-observations" on the underlying scale, X = known coefficient matrix including the incidence of observations in Y related to fixed effects in ß and covariates of heifers’ age, Z = known incidence matrix relating observations in Y to random additive direct genetic effects, ß = p x 1 vector of fixed effects, u = q x 1 vector of random additive direct genetic effects of animal on the underlying scale, e = vector of residual error, A = Wright’s numerator relationship matrix between all animals, I = identity matrix,
= additive direct genetic variance for heifer pregnancy on the underlying scale, and 
= the residual error variance. In this analysis, the residual variance on the underlying scale was constrained to 1 (Gianola and Foulley, 1983).
Fixed effects in the model included the day of year on which the heifer was born as a covariate (representing age of the heifer), contemporary group, and dam’s age class at calving (six classes: 1, up to 27 mo; 2, 28 to 42; 3, 43 to 72; 4, 73 to 120; 5, 121 to 144; and 6, older than 144 mo). Two methods of specifying contemporary groups (CG1 and CG2) were used. There were 45 CG1 heifers, defined as the heifers born in the same herd-year, and exposed to breeding in the same herd-year with the same pasture service sire or multisire group. Service sires were almost always a multisire pasture group and determined the heifer breeding management group. There were 541 CG2 heifers, defined as the heifers born in the same herd-year, weaned in the same management group, and exposed to breeding in the same herd-year with the same pasture service sire or multisire group. Using this definition, 4,168 heifers were in 208 groups with no variation and were removed. From the 13,977 heifers remaining, 10,928 were scored as 0 (78%), and 3,049 were scored as 1 (22%). From the heifers removed, 4,155 (99.7%) were scored as 0, and only 13 (0.3%) were scored as 1.
Random effects included animal additive genetic effect and residual error. The Animal Breeder’s ToolKit software (Golden et al., 1992) was used for variance component estimation and to assemble and solve mixed-model equations. Furthermore, 90% confidence intervals of heritability estimates were generated using a procedure described by Mallinckrodt et al. (1997).
Scrotal Circumference Data
Data on scrotal circumference were obtained from 25,466 young bulls born from 1993 through 2001 in the same herds as heifers. Scrotal circumference was measured at about 18 mo of age up to year 1999. After that, all measurement was done at 15 mo. Data on scrotal circumference were adjusted to 450 d using a segmented polynomial technique (Gallant and Fuller, 1973). This procedure gave a better adjustment curve in previous unpublished analysis. The average adjusted scrotal circumference was 24.1 cm. There were 693 CG, defined as bulls born in the same herd-year and evaluated on the same weaning and postweaning management groups. The young bulls were sons of 446 sires and 16,320 dams. Owing to the use of multisire pastures, 4,288 bulls came from those groups and had unknown sires.
Two-Trait Analyses
Heifer pregnancy and scrotal circumference data were analyzed simultaneously using a two-trait analysis. Pedigree data included all animals with an observation on heifer pregnancy or scrotal circumference plus all known relationships up to nine generations. The pedigree included 75,461 animals, including 966 sires, 33,885 dams and 33,684 base animals.
Statistical Procedures.
Variance components were estimated using Method procedures with a continuous and categorical animal model (Reverter et al., 1994; Kaiser, 1996). The model used for analysis of scrotal circumference (Trait 1) and heifer pregnancy (Trait 2) together can be described as follows (Evans et al., 1999). For scrotal circumference the model was
 |
and for heifer pregnancy was
 |
where
 |
and y1 = observation on scrotal circumference, y2 = observation on heifer pregnancy on the liability scale, X1(2) = known coefficient matrix including the incidence matrix relating fixed effects to observations on Trait 1(2) and age (day of year born) as covariates, Z1(2) = incidence matrix relating random animal additive direct genetic effects to observations on Trait 1(2), b1(2) = vector of fixed effects related to Trait 1(2), u1(2) = vector of random additive direct genetic effects for Trait 1(2), e1 = vector of random errors associated with Trait 1, E(e2 | e1) = expectation of errors associated with Trait 2 given the observed errors for Trait 1, 
= vector of random errors associated with Trait 2 independent of the random errors for Trait 1, and b1 = partial regression coefficient for e2 on e1 (Kaiser and Golden, 1995; Kaiser, 1996). Because scrotal circumference and heifer pregnancy were observed on separate animals, a zero value was assigned to b1.
The matrix g = 2 x 2 additive direct genetic (co)variance, A = Wright’s numerator relationship matrix, 
= Kronecker product, I1(2) = identity matrix of the same order as e1(2), and 
2e1 = variance of the scrotal circumference residual error. The matrix corresponding to the continuous-threshold residual covariance was set to zero because the traits were measured on different animals (Kaiser and Golden, 1995; Kaiser, 1996).
Model and Fixed Effects.
The mathematical model for heifer pregnancy included as fixed effects contemporary groups (45 CG1 or 333 CG2), dam age classes (six classes), and age of the heifer (day of year born) as a linear covariate, and for scrotal circumference, 693 CG were include in the model. The random effects included animal additive genetic effect and residual error.
Genetic Evaluation for Heifer Pregnancy
Breeding values were predicted on the underlying scale using a maximum a posteriori probit threshold model (MAP; Gianola and Foulley, 1983; Harville and Mee, 1984; Evans et al., 1999). Solutions were obtained for all animals in the pedigree file and EPD were computed by transforming MAP solutions for deviations from 50% probability according to the equation EPDi = [
(MAPi x 0.5) – 0.5] x 100 (Snelling et al., 1995), where EPDi = expected progeny difference for the ith animal on the probability scale; = standard accumulated distribution function; and MAPi = solution for the ith animal in the underlying scale.
|
Results and Discussion |
|---|
|
TopAbstractIntroductionMaterial and MethodsResults and DiscussionImplicationsLiterature Cited |
|---|
Solutions for dam’s age class are presented in Table 1
. Heifers from the youngest classes (1, 2, and 3) and the oldest (5 and 6) were less likely to conceive than those from mature dams (Class 4), probably owing to the milking ability of mature dams compared with that of the younger and older ones. The differences are, however, not important except for Class 2 (3-yr-old) dams. It is important to note that heifers from Class 1 (2-yr-old) dams are more likely to conceive than those from Class 2. There is no reason for 2-yr-old dams to milk better than 3-yr-old dams (Class 2). Probably some confounding between genetic and nongenetic effects must be occurring. The majority of Class 2 dams were composed of heifers that did not conceive in the preceding breeding year and these dams could be genetically inferior. As heritability is high, heifers from precocious (Class 1) dams should be genetically better. Heifers from Class 3 dams (4 and 5 yr old) are also supposed to be genetically superior to Class 2 dams because all 3-yr-old heifers that did not calve were culled.
Heritability and Genetic Correlation
Single-Trait Analysis.
The mean heritability estimates for heifer pregnancy using single-trait analyses (Table 2) were 0.68 ± 0.09 and 0.61 ± 0.10 for CG1 and CG2, respectively. Besides higher heritability, using CG1 made convergence easier. The values are the same if the median is taken as a criterion for heritability estimation. The distribution of heritability estimates is presented in Figure 1. Eler et al. (2002) reported an estimated heritability of 0.57 ± 0.11 using a contemporary group definition that included weaning management group of the heifer (CG2) and records of heifers born from 1993 to 1999. Previous studies using the same methodology (Evans et al., 1999; Doyle et al., 2000) reported lower heritability for Hereford and Angus breed, respectively. However, the results reported by Evans et al. (1999) and by Doyle et al. (2000) were obtained for populations of Bos taurus origin for which almost all animals reach puberty at yearling age (14 mo or before that age). The Nellore breed (Bos indicus) has, on average, later ages at puberty, and not all heifers reach puberty at that age. Because puberty is later, genetic variability of heifer pregnancy seems to be much higher. The records show, for example, a bull whose daughters had less than 1% pregnancy rate (1 pregnant out of 110 daughters) and another bull whose daughters had a 55% pregnancy rate (66 out of 120 daughters). Because the daughters of both sires were in some of the same contemporary groups, the difference between them should be in great part genetic.
|
, supporting the use of heifer pregnancy data to develop expected progeny difference for heifer pregnancy. Heifer pregnancy records seem to be better than age at first calving records as a selection criterion for sexual precocity due to high heritability and there is no need to wait for calving record. Data on heifer pregnancy can be analyzed just after pregnancy diagnosis.
For scrotal circumference, single-trait analysis to estimate heritability was not performed because previous work, using maximum likelihood methods (Eler et al., 1996; Quirino and Bergman, 1998; Eler et al., 2001), indicates that scrotal circumference is a heritable trait.
Two-Trait Analysis.
Heritability and genetic correlation estimates from two-trait analyses are presented in Tables 3 and 4, and the distributions of estimates are presented in Figures 2, 3, and 4. The mean heritability estimates for heifer pregnancy were similar to those obtained from single-trait analysis. The analysis using CG2 took much longer to converge because this contemporary group definition leads to many groups with small number of heifers.
|
|
|
|
|
and by Eler et al. (2001) in analyses using maximum likelihood methods and a two-trait model including scrotal circumference and weaning weight.
The genetic correlation estimates were 0.20 ± 0.12 and 0.20 ± 0.13 (Table 4), respectively, when using CG1 and CG2. The estimate of 0.20 is consistent with the results reported by Martínez-Velázquez et al. (2003), who analyzed data from Bos taurus breeds. These results indicate that response to selection for heifer precocity based on scrotal circumference would be much slower than previously believed for Nellore cattle. However, this correlation would be high enough to encourage the incorporation of scrotal circumference in a two-trait analysis in order to increase accuracy of prediction for heifer pregnancy in young bulls.
Results from Genetic Evaluation
Heifer pregnancy EPD was predicted for all animals using a single-trait model and a two-trait animal model and previously reported multivariate genetic parameters. Predictions were presented as the probability of a bull having daughters that will become pregnant when exposed at 14 mo of age during a given breeding season. The following example explains the interpretation of the heifer pregnancy EPD. Two bulls, Sire A and B, are each bred to the same herd. For simplicity, assume that all the progeny are heifers, all are retained for breeding, and all have an equal opportunity for exposure to the same bull. Bull A has a heifer pregnancy EPD of 20 and Bull B an EPD of 0. On average, Bull A’s heifer progeny will have a 20% higher probability of conceiving and remaining pregnant to palpation when compared with heifers from Bull B.
The distribution of EPD is reported in Figure 5 for four groups (all animals, sires, young bulls, and heifers). The correlation between single-trait and two-trait predictions was very high. The values of rank correlation were 0.94, 0.97, 0.91, and 0.98 for all animals, sires, young bulls, and heifers, respectively.
|
For young bulls, despite the high correlation, there was a noticeable change on EPD from single-trait to two-trait analyses, according to the genetic merit of the young bulls for scrotal circumference (Table 5). The scrotal circumference EPD shown in Table 5, were predicted from a BLUP analysis using a two-trait animal model including scrotal circumference and weaning weight records.
|
|
Implications |
|---|
|
TopAbstractIntroductionMaterial and MethodsResults and DiscussionImplicationsLiterature Cited |
|---|
|
Footnotes |
|---|
2 Correspondence and present address: Faculdade de Zootecnia e Engenharia de Alimentos/USP (phone: 55 19 35654074; fax: 55 19 35618606; e-mail: joapeler{at}usp.br).
Received for publication December 4, 2003. Accepted for publication May 10, 2004.
|
Literature Cited |
|---|
|
TopAbstractIntroductionMaterial and MethodsResults and DiscussionImplicationsLiterature Cited |
|---|
estimates. J. Anim. Sci. 74:2041–2046.
variance components procedure: Application of the simple breeding value model. J. Anim. Sci. 72:2247–2253.[Abstract]
This article has been cited by other articles:
|
|
 |
|
  M. H. Van Melis, J. P. Eler, H. N. Oliveira, G. J. M. Rosa, J. A. V. Silva II, J. B. S. Ferraz, and E. Pereira Study of stayability in Nellore cows using a threshold model J Anim Sci, July 1, 2007; 85(7): 1780 - 1786. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. L. L. Guerra, D. E. Franke, and D. C. Blouin Genetic parameters for calving rate and calf survival from linear, threshold, and logistic models in a multibreed beef cattle population J Anim Sci, December 1, 2006; 84(12): 3197 - 3203. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
